The mean age for the group was 35.2 yr (95% C.I.: 30.5 to 39.9) years. There were 2 females, 5 African-Americans, 3 Caucasians, and 1 Caucasian Hispanic (TABLE 1). These subjects were a very normal and healthy group based upon their histories and the strict exclusion criteria.

Mean NPY_CSF was 121.9 (87.8 to 156.0) pg/ml. NPY did not correlate with any variable.

Significant covariates of CRH_CSF could be grouped into: metabolic, autonomic function, and perceptional and cognitive functions (TABLE 3). Serum glucose was positively, and sodium negatively, correlated with CRH_CSF (FIGURE 1). Explained variances (R²) were 0.97 and 0.66, respectively. Glucose and sodium were also negatively correlated (R² = 0.85). Resting norepinephrine levels at 2 time points and heart rate at 4 time points were colinear and positively correlated to CRH_CSF.

CRH_CSF was negatively correlated with the Holter monitor-derived measure of log heart rate summed for the daytime, and the threshold temperature causing an initial, mild sensation of burning pain (Stressor I). The latter indicated that subjects with higher CRH_CSF perceived the burning pain of the cutaneous forearm hyperthermic stimulation at a lower threshold temperature than their peers. This may indicate an increased sensitivity to nociceptive stimuli. However, there was no correlation with deep pressure – induced pain. Negative correlations were also found with the SF-36 Vitality and SES Manage Symptoms domains. High scores were normal for these questionnaires, with lower scores indicating dysfunction.

All parts of the study were completed by at least 5 males. Analysis of the male subgroup gave some information about the role of gender. The pattern of significant covariables was different from the total group (TABLE 4). Respiratory rate was the only vital or physiological sign related to CRH_CSF. Perceptions of vulnerability, physical functioning and self-efficacy were more highly related.

Sets of the significant metabolic, autonomic, and perceptual variables were grouped and analyzed by 3 linear regression models in order to detect significant relationships between independent variables and CRH_CSF. The 3 models had high significance levels (TABLE 5). A metabolic model related CRH_CSF to glucose and sodium. An autonomic model linked CRH_CSF to resting respiratory and heart rates. The optimum perceptual model predicted CRH_CSF based on SF-36 Vitality and MFI Physical Fatigue. The latter 2 domains were not the same as those in TABLES 3 and 4 because many of these variables were co-linear or surrogates of one another. This was reinforced by the similarity of R² and p values where these domains were substituted for Vitality and Physical Fatigue. These results were remarkable because it is very unusual for models with such small numbers of observations (n ≤ 9) to be so significant (p < 0.02). The explained variances were very high (R² > 0.85) suggesting that the factors may be causally connected.

CRH_CSF did not correlate with variables associated with maximum exercise, heat – and pressure – (dolorimetry) induced pain, or other rapid onset stressors (TABLE 1).

Nine normal, healthy subjects (2 females) gave informed consent for this paid, Institutional Review Board – approved protocol. They had a comprehensive screening evaluation to exclude: severe physical impairment, morbid obesity, autoimmune/inflammatory diseases, cardiopulmonary disorders, uncontrolled endocrine or allergic disorders, malignancy, severe psychiatric illnesses (e.g., schizophrenia, substance abuse), factors known to affect the HPA axis or autonomic function (cigarette smoking, daily intake of caffeine exceeding the equivalent of 2 cups of coffee), or medication use. Subjects had a history and physical examination, were administered the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) (SCID II) 1920, Composite International Diagnostic Interview (CIDI) 2122 and Center for Epidemiologic Studies Depression scale (CESD) 23 to detect psychiatric co-morbidities, and completed the self – report Short Form 36 (SF-36) 2425, Self Efficacy Scale (SES) 26, Meaning of Illness Questionnaire (MIQ) 27, and Multidimensional Fatigue Inventory (MFI) 28.

Subjects were admitted to the G-CRC on the evening of Day 1. An 18-gauge catheter was inserted antecubitally and infused with normal saline at 50 ml/hr. A Holter monitor was attached to monitor autonomic regulation of cardiac rhythm 293031. Upon awakening, or at 6:30 am, subjects stayed in bed until they had "pre-awakening" blood tests drawn. Breakfast was served between 7:00 and 8:00 a.m. and baseline Day 2 blood samples drawn at 8:30 a.m. Vital signs, questionnaire responses, and other blood samples were drawn from the catheter to assess HPA axis and stress system responses before, during and after Day 2 stressors. Pain responses were tested by pressure applied to each subject's thumbnail and heat to the forearm applied in random staircase testing paradigms (Stressor I) 3233. Stressor II was a series of cognitive challenges including the Benton Visual Retention Test of visual-spatial memory; Digit Span recitation; and Pig Latin, a test of verbal working memory 34. Lunch was served, followed by a 2 to 2.5 hours rest break to allow post-prandial catecholamine levels to reequilibrate. At 2 pm, serum glucose, electrolytes, plasma catecholamines and other analytes were measured. Subjects then squeezed an isometric hand grip dynamometer to test autonomic function and muscular fatigue 35 (Jamar, Sammons Preston, Bollingbrook, IL) (Stressor III). After 30 min of rest, they had a sub-maximal exercise test on an electronically-braked cycle ergometer (Sensormedics, Yorba Linda, CA). The test was graded in 3 min stages and ended when the subjects' heart rates reached 85% of their age-predicted maximum (Stressor IV). Lumbar punctures were performed 30 min later (approximately 4 pm) (Stressor V). Subjects sat while sterile technique was used to prepare the skin over the L4–L5 lumbar region, infiltrate the subcutaneous and deep tissues with 2% lidocaine, and insert a 22G spinal catheter. CSF was collected as 3 to 4 aliquots of about 2 ml each. Catheters were withdrawn and subjects allowed to rest in their preferred position for 30 min.

Tubes of CSF were immediately placed on ice and then centrifuged at 4°C. The supernatants were rapidly frozen at -80°C. Tube 2 or 3 was removed from the freezer and thawed at 4°C. Peptides were extracted by precipitating high molecular weight proteins by adding an equal volume of 100% ethanol, 0.1 M acetic acid, 0.2% sodium bisulfite 3637. The supernatant was dried (SpeedVac, Thermo Savant, Holbrook, NY), resuspended in phosphate buffered saline with 1% bovine serum albumin (RIA buffer; Peninsula Laboratories, Inc., San Carlos, CA.). Samples and standard amounts of each peptide were aliquoted and peptide specific rabbit antibodies added. After overnight incubation at 4°C, I¹²⁵-CRH or -NPY was added, tubes gently vortexed, and again incubated overnight. Goat anti-rabbit antibodies were added, tubes incubated, and immune complexes precipitated by centrifugation. Radioactivity was counted for the standards, and the concentrations for each sample interpolated from the standard curves. Spiking CSF with fixed amounts of I¹²⁵-peptides and performing the RIA shifted the curves to the left by the anticipated concentrations. Standard curves were reproducible to within 10%.

Plasma catecholamine levels were measured by HPLC (Mayo Clinic Laboratories, Rochester, MN).

All the data from the provocation studies, questionnaire domains, blood work, and neuropeptide analysis were entered into a SAS spreadsheet (SAS, Carey, NC) using sequential hand or scanner entry followed by data checking routines.

Means with 95% confidence intervals were reported.

Simple correlations between our outcome measure (CRH_CSF) and other objective and subjective patient variables were conducted as an exploratory tool. Given the small sample size (n = 9), both parametric and non-parametric correlations were used to evaluate the robustness of the correlation coefficients. Partial and intra-class correlations examined the structure of the relationships between CRH_CSF and the independent variables. Original data were always checked to make sure that correlations were not spurious, due to outliers, colinearity, or sets of virtually identical scores.

Our data included a very wide range of covariates which spanned the physical and psychological attributes of the patients. Scatter plots were used to determine the response slopes with CRH_CSF. Independent variables with flat slopes or high scatter that did not generate correlations with CRH_CSF, that were highly correlated with each other (collinear), or that had < 5 recorded values were excluded from further analysis.

The remaining suitable independent variables fell into 3 categories: metabolic, autonomic and psychometric function. Three 3 separate linear regression models 8 were constructed to predict CRH_CSF, the dependent variable. Separate models were required because of the low degrees of freedom available for the analyses, and to ensure that the explained variance (R²) was not inflated due to overloading the model with potentially collinear variables. Variance inflation factors and tolerance values were incorporated to optimize the selection of the most independent combinations of variables that also maximized the R² and p values. The linear regression procedure took into account the multiple comparisons for each model.